0195 Integrated actigraphy-based biomarker for the risk of Alzheimer's dementia. (25th May 2022)

• Record Type: Journal Article
• Authors: Yang, Hui-Wen; Li, Peng; Sun, Haoqi; Maher, Matthew; Lane, Jacqueline; Lim, Andrew; Bennett, David; Yu, Lei; Saxena, Richa; Buchman, Aron; Hu, Kun
• Is Part Of: Sleep. Volume 45(2022)Supplement 1
• Page Start: A89
• Page End: A89
• Publication Date: 2022-05-25
• Subjects: Sleep -- Physiological aspects -- Periodicals; Sleep disorders -- Periodicals; Sommeil -- Aspect physiologique -- Périodiques; Sommeil, Troubles du -- Périodiques; Sleep disorders; Sleep -- Physiological aspects; Sleep -- physiological aspects; Sleep Wake Disorders; Psychophysiology; Electronic journals; Periodicals; 616.8498
• DOI: 10.1093/sleep/zsac079.193
• Languages: English
• ISSNs: 0161-8105

Abstract

Abstract: Introduction: Many physiological measures derived from actigraphy including physical activity, sleep, circadian/daily rhythm, and temporal correlations have been shown to predict Alzheimer's dementia (AD). This study aimed to combine these actigraphy-based measures to develop an integrated actigraphy biomarker (IAB) for AD and to test its link to the genetic risk for AD. Methods: We analyzed data of 1107 participants (age 80.9±7.3(mean±SD)) from the Rush Memory and Aging Project who were non-demented and had actigraphy (~10 days) at baseline, and had annual cognitive assessment during the follow-up (1-15 years). 270 developed AD (mean = 7.4 years). To construct the IAB for the AD's risk, we trained a random forest survival model, in which time to incident AD was the outcome, and inputs included 10 features derived from actigraphy data: physical activity level, 3 features for sleep (sleep duration, sleep fragmentation, activity fragmentation), 4 features for circadian rhythmicity (amplitude, acrophase, interdaily stability, and intradaily variability of 24-hr rhythms), and 2 features for temporal correlations (at timescales between 1-90 min and 120-480 min). Polygenic risk score (PRS) was calculated using 457 independent SNPs strongly associated with Alzheimer's disease (p<0.001). Cox proportional hazard ratio models were performed with different combinations of IAB, PRS, age, sex, and education, and the concordance score (C-score) was used to evaluate model performance. Results: The derived IAB was 0.6 SD larger in the AD group as compared with the controls. The IAB alone achieved a C-score = 0.61 in predicting AD, with a hazard ratio=1.5 for 1-SD increase in IAB. The IAB and PRS were not correlated (r2=0.0004, p=0.25), and both significantly contributed to the prediction (both p<=0.0001) when included in one model, giving a C-score of 0.65. C-score was 0.7 in the model using only age, sex and educations yielded, and increased to 0.74 after including IAB and PRS (both effects remained significant p<0.0001). Conclusion: The integrated actigraphy biomarker may provide complementary information for early prediction and detection of AD, independent of the known demographic and genetic risk factors. Support (If Any): NIH (RF1AG064312, RF1AG059867, R01AG56352, R01AG17917, T32GM007592, and R03AG067985); The BrightFocus Foundation (A2020886S).